3

Basic Thermal Design Theory for Recuperators

As defined in Chapter 1, in a recuperator, two fluids are separated by a heat transfer surface (wall), these fluids ideally do not mix, and there are no moving parts. In this chapter the thermal design theory of recuperators is presented. In a heat exchanger, when hot and cold fluids are maintained at constant temperatures of Th and T, as shown in Fig. 3.la, the driving force for overall heat transfer in the exchanger, referred to as mean temperature dtyerence (MTD), is simply Th - T,. Such idealized constant temperatures on both sides may occur in idealized single-component condensation on one fluid side and idealized single-component evaporation on the other fluid side of the exchanger. However, a number of heat transfer applications have condensation or evaporation of single-component fluid on one side and single-phase fluid on the other side. In such cases, the idealized temperature distribution is shown in Fig. 3.lb and c. The mean temperature difference for these cases is not simply the difference between the constant temperature and the arithmetic mean of the variable temperature. It is more complicated, as will be discussed. In reality, when the temperatures of both fluids are changing during their passage through the exchanger (see, e.g., Figs. 1.50, 1.52, 1.54, 1.56b, 1.57b, 1.60b, 1.63b and 1.646), the determination of the MTD is complex. Our objective in this chapter is to conduct the appropriate heat transfer analysis in the exchanger for the evaluation of MTD and/or performance. Subsequently, design methods are outlined and design problems will be formulated. The following are the contents of this chapter: An analogy between thermal, fluid, and electrical parameters is presented in Section 3.1. Heat exchanger variables and the thermal circuit are presented in Section 3.2. The E-NTU method is introduced in Section 3.3. Specific E-NTU relationships for various flow arrangements are summarized in Section 3.4. The P-NTU method is introduced in Section 3.5, and P-NTU relationships for various flow arrangements are summarized in Section 3.6. The mean temperature difference (MTD) method is introduced in Section 3.7. The MTD correction factors F for various flow arrangements are presented in Section 3.8. It is shown in Section 3.9 that the results of applications of E-NTU and MTD methods are identical, although each method has some limitations. The $-P and P1-Pz graphical presentation methods, which eliminate some of the limitations of the aforementioned methods, are presented in Section 3.10. A brief description of various methods used to obtain E-NTU or P-NTU formulas for various exchanger flow arrangements is presented in Section 3.1 1. Considering seven variables of the heat exchanger design problem, there are a total of 21 design problems possible, as discussed in Section 3.12.
9 7

98

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

0 SurfaceareaA

SurfaceareaA

o SurfaceareaA L

(4

(b)

(4

FIGURE 3.1 Idealized temperature distributions of one or both single-component phase-change fluids: (a) one fluid condensing, the other evaporating; ( b ) one single-phase fluid cooling, the other fluid evaporating; (c) one fluid condensing, the other single-phase fluid heating.

3.1 FORMAL ANALOGY BETWEEN THERMAL AND ELECTRICAL ENTITIES

In heat exchanger analysis, a formal analogy between heat transfer and conduction of electricity is very useful; to understand this analogy clearly, let us start with the definitions. Heat Row q is a consequence of thermal energy in transit by virtue o f a temperature difference A T . By Ohms law, the electric current i is a consequence of electrical energy in transit by virtue of an electromotive force (potential) difference AE. In both cases, the rate of Row of related entity is inhibited by one or more recognizable resistances acting in series and/or in parallel combinations. Heat Flow (Heat Transfer Rate) Electric Current Flow

With this notion, theformal analogy between various parameters is presented in Table 3.1. It is important to note that the relationships between current, potential, resistance, conductance, capacitance, and time constant terms are analogous for these different physical processes. While the expressions for power and energy are analogous for heat and current flow from the physics point of view, they are not analogous from the resistance circuit point of view as their formulas differ as shown in Table 3.1. Moreover, there is no thermal analogy to electrical inductance or inertia in this analogy (not shown in Table 3.1). Note that heat capacity or thermal capacitance energy storage terminology used in heat transfer is used incorrectly as thermal inertia in the literature. Since we know electrical circuit symbolism, we will find it convenient to borrow the symbols for the thermal circuits used to describe the exchanger heat transfer process. These are summarized in Fig. 3.2. We will also need an analogy between fluid Row in a pipe (exchanger) and electric current for the pressure drop analysis of the exchanger. The basic parameters of pressure drop (head), fluid flow rate, and Row losses are analogous to the voltage potential,

FORMAL ANALOGY BETWEEN THERMAL AND ELECTRICAL ENTITIES

99

TABLE 3.1 Analogies and Nonalogies between Thermal and Electrical Parameters Parameter Current Potential Resistance Conductance Capacitance Time constant Power Energy Electrical Analogies ampere, A volts, v ohms, R, V/A siemens, S, A,'V farads, F, A sjV
S

Thermal
4 AT R = I/UA UA W, Btu/hr "C(K),"F("R) "C/W, OF-hr/Btu W/T, Btu/hr-"F W s/"C,Btu/"F s, hr W, Btu/hr

E R G C RC

RC Nonanalogies
4

iE

J, Btu

Resistance Capacitance Potentialsource (constant) Potential soume (periodic) Current source (constant) Current s o u m (periodic)

FIGURE 3.2 Thermal circuit symbolism.

current, and resistance.+Since the flow losses are measured in terms of the pressure loss or head o r fluid column, which have the same units as the potential, this analogy is not as well defined as the analogy between heat transfer and electric current. Again, the relationship between analogous parameters for fluid flow is not linear for transition and turbulent flows o r developing laminar flows in pipes.

' Pipe and duct design based on one-dimensional lumped parameter analysis typically defines the flow resistance or
the f l o w loss coefficient K as equal to the number of velocity heads lost due to frictional effects [see Eq. (6. 53)].

100

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

3.2 HEAT EXCHANGER VARIABLES AND THERMAL CIRCUIT

In this section, starting with the assumptions built into heat exchanger design theory, the basic problem for the exchanger heat transfer analysis is formulated. This includes the differential equations used for the analysis as well as a list of independent and dependent variables associated with heat exchanger design and analysis problems. Next, the basic definitions of important dimensional variables and important terminologies are introduced. Finally, the thermal circuit, and expressions for CIA and wall temperatures are presented.

3.2.1 Assumptions for Heat Transfer Analysis

T o analyze the exchanger heat transfer problem, a set of assumptions are introduced so that the resulting theoretical models are simple enough for the analysis. The following assumptions and/or idealizations are made for the exchanger heat transfer problem formulations: the energy balances, rate equations, boundary conditions, and subsequent analysis [see, e.g., Eqs. (3.2) and (3.4) through (3.6) in differential or integral form].+ 1. The heat exchanger operates under steady-state conditions [i.e., constant flow rates and fluid temperatures (at the inlet and within the exchanger) independent of time]. 2. Heat losses to or from the surroundings are negligible (i.e. the heat exchanger outside walls are adiabatic). 3. There are no thermal energy sources or sinks in the exchanger walls or fluids, such as electric heating, chemical reaction, or nuclear processes. 4. The temperature of each fluid is uniform over every cross section in counterflow and parallelflow exchangers (i.e., perfect transverse mixing and no temperature gradient normal to the flow direction). Each fluid is considered mixed or unmixed from the temperature distribution viewpoint at every cross section in single-pass crossflow exchangers, depending on the specifications. For a multipass exchanger, the foregoing statements apply to each pass depending on the basic flow arrangement of the passes; the fluid is considered mixed or unmixed between passes as specified. 5. Wall thermal resistance is distributed uniformly in the entire exchanger. 6. Either there are no phase changes (condensation or vaporization) in the fluid streams flowing through the exchanger or the phase change occurs under the following condition. The phase change occurs at a constant temperature as for a single-component fluid at constant pressure; the effective specific heat c ~for , ~ f fm, the phase-changing fluid is infinity in this case, and hence C,,, = r i Z ~ ~ , + where riZ is the fluid mass flow rate. 7. Longitudinal heat conduction in the fluids and in the wall is negligible. 8. The individual and overall heat transfer coefficients are constant (independent of temperature, time, and position) throughout the exchanger, including the case of phase-changing fluids in assumption 6.
The complete set of differential equations and boundary conditions describing the mathematical models of heat exchangers, based on these assumptions, is presented in Section 11.2.

'

HEAT EXCHANGER VARIABLES A N D THERMAL CIRCUIT

101

9. The specific heat of each fluid is constant throughout the exchanger, so that the heat capacity rate on each side is treated as constant. Note that the other fluid properties are not involved directly in the energy balance and rate equations, but are involved implicitly in NTU and are treated as constant. 10. For an extended surface exchanger, the overall extended surface efficiency qo is considered uniform and constant. 11. The heat transfer surface area A is distributed uniformly on each fluid side in a single-pass or multipass exchanger. In a multipass unit, the heat transfer surface area is distributed uniformly in each pass, although different passes can have different surface areas. 12. For a plate-baffled 1-n shell-and-tube exchanger, the temperature rise (or drop) per baffle pass (or compartment) is small compared to the total temperature rise (or drop) of the shell fluid in the exchanger, so that the shell fluid can be treated as mixed at any cross section. This implies that the number of baffles is large in the exchanger. 13. The velocity and temperature at the entrance of the heat exchanger on each fluid side are uniform over the flow cross section. There is no gross flow maldistribution at the inlet. 14. The fluid flow rate is uniformly distributed through the exchanger on each fluid side in each pass i.e., no passage-to-passage or viscosity-induced maldistribution occurs in the exchanger core. Also, no flow stratification, flow bypassing, or flow leakages occur in any stream. The flow condition is characterized by the bulk (or mean) velocity at any cross section.

Assumptions 1 through 5 are necessary in a theoretical analysis of steady-state heat exchangers. Heat losses to the surroundings, if small, may be taken into account approximately by using the effective heat capacity rate Cefffor the hot fluid instead of the actual C (= h c P )in the analysis. Cen is determined based on the actual heat transfer rate from the hot to cold fluid. Assumption 6 essentially restricts the analysis to single-phase flow on both sides or one side with a dominating thermal resistance. For two-phase flows on both sides, many of the foregoing assumptions are not valid since mass transfer in phase change results in variable properties and variable flow rates of each phase, and the heat transfer coefficients may also vary significantly. As a result, the E-NTU and other methods presented in Sections 3.3 through 3.11 are not applicable to two-phase heat exchangers. Assumptions 7 through 12 are relaxed in Chapter 4. Assumptions 13 and 14 are addressed in Chapter 12. If any of the foregoing assumptions are not valid for a particular exchanger application and the sections that cover the relaxation of these assumptions do not provide a satisfactory solution, the approach recommended is to work directly with Eqs. (3.3) and (3.4), or a set of equations corresponding to the model. In this case, modify these differential equations by including a particular effect, and integrate them numerically across sufficiently small segments of the exchanger in which all the assumptions are valid. Refer to Section 4.2.3.2 for an example. In Sections 3.3 through 3.11, we present E-NTU, P-NTU, MTD, @-P,and P 1 - P 2 methods of exchanger heat transfer analysis for which the 14 assumptions are invoked. The corresponding model building, based on these assumptions, is discussed in detail in Section 11.2.

104

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

3.2.2 Problem Formulation

T o perform the heat transfer analysis of an exchanger, our objective is to relate the heat transfer rate q, heat transfer surface area A , heat capacity rate C of each fluid, overall heat transfer coefficient CJ, and fluid terminal temperatures. Two basic relationships are used for this purpose: (1) energy balance based on the first law of thermodynamics, and (2) rate equations for heat transfer, as outlined by Eqs. (2.1) and (2.2). Consider a two-fluid exchanger (a counterflow exchanger as an example) shown in Fig 3.3 to arrive at the variables relating to the thermal performance of a two-fluid exchanger. Schematic of Fig. 3.3 and the balance equations for different exchanger flow arrangements may be different, but the basic concept of modeling remains the same. The analysis that will follow is intended to introduce variables important for heat exchanger analysis. Detailed approaches to a general thermodynamic problem formulation are presented in Chapter 11. Two differential energy conservation (or balance) equations (based on the energy balance implied by the first law of thermodynamics) can be combined as follows for control volumes associated with the differential element of dA area for steady-state flow, an overall adiabatic system, and negligible potential and kinetic energy changes:

Heat transfer area A

- x

FIGURE 3.3 Nomenclature for heat exchanger variables (From Shah, 1983).

HEAT EXCHANGER VARIABLES AND THERMAL CIRCUIT

103

The negative signs in this equation are a result of Th and T, decreasing with increasing A (these temperatures decrease with increasing flow length x as shown in Fig. 1 .SO);+ also, dq is the heat transfer rate from the hot to cold fluid, C = mcp is the heat capacity rate of the fluid, m is the fluid flow rate, c,, is the fluid specific heat at constant pressure, T is the fluid temperature, and the subscripts h and c denote hot and cold fluids, respectively. The heat capacity rate C J p . "C (Btu/hr. O F ) is the amount of heat in joules (Btu) that must be added to or extracted from the fluid stream per second (hour) to change its temperature by 1C ( O F ) . The product kc,, = C appears in the energy balance [Eq. (3.2)] for constant c,,, and hence C is commonly used in the heat exchanger analysis instead of m and cp as two parameters separately. In general, for any isobaric change of state, Eq. (3.2) should be replaced by

where h is the fluid specific enthalpy, J/kg (Btu/lbm). If the change of state is a phase change, enthalpy differences should be replaced by enthalpies of phase change (either enthalpy of evaporation or enthalpy of condensation). However, cp can be assumed as infinity for condensing or evaporating single-component fluid streams. Hence, the phasechanging stream can be treated as a "single-phase" fluid having A T = q/C or dT = d q / C , with C being infinity for a finite q or dq since the A T or dT = 0 for isothermal condensing or evaporating fluid streams (see Fig. 3.1). Note that here A T = Th,i - Th,o or Tc,o- Tc,ias appropriate. The overall heat transfer rate equation on a differential base for the surface area dA of Fig. 3.3 is (3.4) where U is the local overall heat transfer coefficient to be defined in Eq. (3.20).$Thus for this differential element dA, the driving potential for heat transfer is the local temperature difference ( T h- T , ) = A T and the thermal conductance is U d A . Integration of Eqs. (3.2) and (3.4) together over the entire heat exchanger surface for specified inlet temperatures will result in an expression that will relate all important operating variables and geometry parameters of the exchanger. The derivation of such an expression for a counterflow exchanger will be presented in Section 3.3 together with final results for many industrially important exchanger flow arrangements. The common assumptions invoked for the derivation and integration of Eqs. (3.2) and (3.4) are summarized in Section 3.2.1. Two basic equations, energy conservation (balance) and rate equations, could also be written on an overall basis for the entire exchanger as follows (under the conditions implied by the above-mentioned idealizations):
'The sign convention adopted in Eq. (3.2) leads to positive value of heat transfer rate change along each d.x element, and should be considered only as formal (i.e., not necessarily in agreement with thermodynamic convention for heat). Note that although the overall heat transfer coefficient U is idealized as constant (see assumption 8 in Section 3.2.1), it can vary significantly in a heat exchanger. In that case, the mean overall heat transfer coefficient U,,, is obtained from local U data (see Section 4.2.3). Even though U = Un, = constant throughout this chapter, we distinguish between U and Up,in Sections 3.2.3 and 3.2.4 to emphasize how the theory is developed using U and
Urn.

104

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

q=

U A T d A = U,AAT,

(3.6)

Here the subscripts i and o denote inlet and outlet, respectively; Th,o and Tc,o represent the outlet temperatures; they are bulk temperatures defined by Eq. (7.10) for a nonuniform temperature distribution across a cross section; and U, and A T, are the mean overall heat transfer coefficient and the exchanger mean temperature difference, respectively, and will be defined shortly. From Eqs. (3.5) and (3.6) and Fig. 3.3, the steady-state overall-adiabatic heat exchanger behavior can thus be presented in terms of dependent fluid outlet temperatures or heat transfer rate as functions of four operating condition variables and three designer controlled parameters:

- -.
dependent variables

Th,o,Tc,oor q = 4 ( Th,irTc,i, C,, Cc U ,A , flow arrangement) 4

+

(3.7)

operating condition variables

independent variables and parameters

.-

parameters under designers control

This equation represents a total of six independent and one or more dependent variables for a given heat exchanger flow arrangement. Of course, any one of the independent variables/parameters in Eq. (3.7) can be made dependent (if unknown); in that case, one of the three dependent variables in Eq. (3.7) becomes an independent variable/parameter. Thus the most general heat exchanger design problem is to determine any two unknown variables from this set when the rest of them are known. For heat exchanger analysis, it is difficult to understand and work with such a large number of variables and parameters as outlined in Eq. (3.7). From dimensional analysis, three dimensionless groups are formulated from six independent and one or more dependent variables of Eq. (3.7). The reduced number of nondimensional variables and parameters simplifies much of the analysis, provides a clear understanding of performance behavior, and the results can be presented in more compact graphical and tabular forms. The specific form of these groups is to some extent optional. Five such options have been used, depending on which method of heat transfer analysis has been used: the effectiveness-number of heat transfer units (E-NTU or PNTU) method, the mean temperature difference (MTD) method, the nondimensional mean temperature difference-temperature effectiveness (+P) method, and the P I-Pz method. These methods are discussed in Sections 3.3 through 3.10.

3.2.3 Basic Definitions

The definitions of the mean overall heat transfer coefficient and mean temperature difference are introduced first. The rate equation (3.4), after rearrangement, is presented in integral form as

Iqg=sA
U dA

HEAT EXCHANGER VARIABLES A N D THERMAL CIRCUIT

105

Now we define the mean temperature difference and mean overall heat transfer coefficient as follows using the terms in Eq. (3.8):

(3.10) Substitution of Eqs. (3.9) and (3.10) into Eq. (3.8) results into the following equation after rearrangement:
q = U,A AT,

(3.1 1)

Here U, is the mean overall heat transfer coefficient, and AT, is the true (or effective) mean temperature difference (MTD), also referred to as mean temperature driving potential or force for heat transfer. Generally, the overall heat transfer coefficient is treated as constant in the heat exchanger analysis. It is simply designated as U , without subscripts or overbars, throughout the book except for Section 4.2, where various definitions of mean overall heat transfer coefficients are introduced. Thus the overall rate equation (3.6) is simply
q = UA AT,

(3.12)

Note that if U is treated as a constant, integration of Eq. (3.4) will yield (3.13) Other commonly used important entities for heat exchangers are the inlet temperature difference, temperature range, temperature approach, temperature pinch, temperature gap, temperature meet, and temperature cross. They are discussed below and summarized in Table 3.2. The inlet temperature diference (ITD) is the difference between inlet temperatures of the hot and cold fluids and is designated as AT,,,,, in this book. AT,,,,, = Th,i - Tc>i is also sometimes referred to as the temperature span or temperature head in a heat exchanger. The temperature range for a fluid is referred to as its actual temperature rise or drop AT within the exchanger. The temperature ranges for hot and cold fluids in the exchangers are then AT, = Th,i- Ths0 and AT, = T,, - Tc,i, respectively. The temperature approach for exchangers with single-phase fluids is defined as the difference between outlet fluid temperatures ( Th,o - T,,o)for all single-pass and multipass flow arrangements except for the counterflow exchanger. For the latter, it is defined as the smaller of (Th,i- Tc,o) and ( Th,o- Tc,i). The temperature approach for multiphase multicomponent streams is defined as the minimum local temperature difference between hot and cold fluid streams. This could occur anywhere in the exchanger, depending on the flow arrangement, heat duty, and so on. It can be shown that the temperature approach for single-phase exchangers is related to the exchanger effectiveness E defined by Eq. (3.44) later as follows:

106

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

TABLE 3.2 Expressions for Temperature Span, Range, Approach, Pinch, Gap, Meet, and Cross
Item Inlet temperature difference, ITD, maximim temperature span o r temperature head Temperature range for hot fluid Temperature range for cold fluid Temperature approach, counterflow exchanger Temoerature approach, all other exchangers Temperature pinch at a local point in an exchanger Temperature gap Temperature meet, counterflow sigle-phase exchanger Temperature meet, all other single-phase exchangers Temperature cross, single-pass exchangers Temperature cross, multipass exchangers Expression

temperature approach =

{ [(

(1 - E ) 4Tm,, for a counterflow exchanger 1 - ( I + C*)E] AT,,,,, = ( Th,o - T,,o) for other exchangers (3.14)

where AT,,, = T,l,i- Tc,iand C* = Cmin/Cmax. For some shell-and-tube, multipass, and two-phase exchangers, it may not be either easy or possible to determine quantitatively the magnitude of the temperature approach. In that case, while the foregoing definition is valid, it loses its usefulness. A temperature pinch refers to a local temperature difference within an exchanger (or an array of exchangers) that is substantially less than either of two terminal temperature differences and is minimum in the exchanger. In the limit, it can approach zero, which is referred to as temperature meet defined below. The temperature pinch usually occurs in an exchanger with two-phase multicomponent fluids. In reality, a temperature pinch close to zero will require a very large (approaching infinity) surface area for a singlepass exchanger. Hence, for a finite-size heat exchanger, the exchanger would cease to function efficiently beyond the temperature pinch point (i.e., resulting in a more significant reduction in heat transfer than justified). However, for a multipass exchanger, the temperature pinch could occur in one pass, and in that case, the performance of that pass beyond the temperature pinch is reduced significantly. A temperature gap refers to the temperature difference between hot and cold fluid outlet temperatures provided that Th,o> Tc,O. A temperature meet refers to the case when the temperature pinch is zero or the hot and cold fluid temperatures are the same locally somewhere in the exchanger or at outlets. This is an idealized condition and does not occur in a single-pass heat exchanger, but may occur in one of the passes of a multipass exchanger. A temperature cross refers to the case when the cold fluid temperature becomes equal or greater than the hot fluid temperature within the exchanger. External temperature cross refers to the case when the cold fluid outlet temperature Tc,o is greater than the hot fluid outlet temperature Th,o. There is no physical (actual) crossing of hot and cold fluid temperature distributions within an exchanger. This is quite common in a counterflow exchanger (see Fig. 1.50a) and other single-pass and multipass exchangers having high

HEAT EXCHANGER VARIABLES A N D THERMAL CIRCUIT

107

NTUs (see Section 3.3.3 for the definition of NTU). The magnitude of the external temperature cross is T,,, - Th,,. Internal temperature cross refers to the case when locally somewhere within the exchanger T, becomes equal to Th (within a pass or in a segment of the exchanger), and beyond that point along the flow length, T, > Th in that pass or segment; subsequently, reverse heat transfer takes place (original cold fluid transferring heat to the original hot fluid). The external temperature cross can occur with or without the internal temperature cross; similarly an internal temperature cross can occur with or without an external temperature cross (see Section 1 1.4.1).
3.2.4 Thermal Circuit and UA

To understand the exchanger overall heat transfer rate equation [Eq. (3.12)], consider the thermal circuit model of Fig. 3.4. Scale or fouling deposit layers are also shown on each side of the wall. In the steady state, heat is transferred from the hot fluid to the cold fluid by the following processes: convection to the hot fluid wall, conduction through the wall, and subsequent convection from the wall to the cold fluid. In many heat exchangers, a fouling film is formed as a result of accumulation of scale or rust formation, deposits from the fluid, chemical reaction products between the fluid and the wall material, and/or biological growth. This undesired fouling film usually has a low thermal conductivity and can increase the thermal resistance to heat flow from the hot fluid to the cold fluid. This added thermal resistance on individual fluid sides for heat conduction through the fouling film is taken into account by afouling factort rr = l/hf, where the subscriptf denotes fouling (or scale); it is shown in Fig. 3.4. Thus, the heat transfer rate per unit area at any section dx (having surface areas dA;,, dA,, etc.) can be presented by the appropriate convection and conduction rate equations as follows:

Alternatively (3.16) where the overall differential thermal resistance dR, consists of component resistances in series (similar to those shown in Fig. 3.4b for a heat exchanger):
-- - dR, = dRh

U dA

+ dRh,/ + dR,,.+ dR,J + dR,

(3.17)

or

+ W e also refer to the fouling factor as the unif fouling fhermal resistance or concisely as fouling resistance r/ = Rf = l / h / = 6 / / k / where 6 , is the thickness of fouling layer and k, is the thermal conductivity of the fouling material. Refer to Section 13.1 for more details on the fouling resistance concept.

108

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

Hot fluid Scale or fouling on the hot side Wall Scale or fouling on the cold side

FIGURE 3.4 ( a )Thermal resistances; (b)thermal circuit for a heat exchanger (From Shah, 1983).

Various symbols in this equation are defined after Eq. (3.24). If we idealize that the heat transfer surface area is distributed uniformly on each fluid side (see assumption 11 in Section 3.2.1), the ratio of differential area on each fluid side to the total area on the respective side remains the same; that is, (3.19) Replacing differential areas of Eq. (3.18) by using corresponding terms of Eq. (3.19), we get

It should be emphasized that U and all hs in this equation are assumed to be local. Using the overall rate equation [Eq. (3.6)], the total heat transfer rate will be

HEAT EXCHANGER VARIABLES A N D THERMAL CIRCUIT

109

and a counterpart of Eq. (3.15) for the entire exchanger is

where the subscript e denotes the effective value for the exchanger, or ( Th,e - Tc,e) = AT,. To be more precise, all individual temperatures in Eq. (3.22) should also be mean or effective values for respective fluid sides. However, this additional subscript is not included for simplicity. In Eq. (3.21), the overall thermal resistance R , consists of component resistances in series as shown in Fig. 3.4b.
(3.23)

For constant and uniform U and h's throughout the exchanger, Eqs. (3.24) and (3.20) are identical. In that case, U, = U and we will use U throughout the book except for Section 4.2. In Eqs. (3.20) and (3.24), depending on the local or mean value, we define
1 1 or (%hA)h (%hmA)h 1 1 or Rh,f = hot-fluid-side fouling resistance = (%hfA)h ('%hm,fA)h

Rh = hot-fluid-side convection resistance =

R , = wall thermal resistance expressed by Eq. (3.25) or (3.26) R,,/ = cold-fluid-side fouling resistance =
~

R , = cold-fluid-side convection resistance = -or (qohA)c (qohmA)c

~

1 1 or (qohfA)c (qohrn,fA)c 1 1

In the foregoing definitions, h is the heat transfer coefficient, discussed in detail in Section 7.1.4.3;h/ is referred to as the fouling coefficient (inverse of fouling factor), discussed in Chapter 13; A represents the total of primary and secondary (finned) surface area; and q,, is the extended surface efficiency of an extended (fin) surface defined in Section 4.3.4. In the literature, l/(qoh)f= R f A = Rf is referred to as unitfouling resistance. Note that no fins are shown in the upper sketch of Fig. 3.4; however, q, is included in the aforementioned various resistance terms in order to make them most general. For all prime surface exchangers (i.e., having no fins or extended surface), qO,hand qo,,are unity. The wall thermal resistance R,. for flat walls is given by

R, =

[&

for flat walls with a single layer wall

(3.25)

for flat walls with a multiple-layer wall

110

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

For a cylindrical (tubular) wall, it is given as for N , circular tubes with a single-layer wall

4, =

i1
27rLN,

n(4+dJ)
k,,.J

for N , circular tubes with a multiple-layer wall (3.26)

where ~ 5 ,is ~ the wall plate thickness, A l v the total wall area of all flat walls for heat conduction, k,v the thermal conductivity of the wall material, do and dithe tube outside and inside diameters, L the tube or exchanger length, and N , the number of tubes. A flat (or plain) wall is generally associated with a plate-fin or an all-prime-surface plate exchanger. In this case, (3.27) Here L , , L2, and N p are the length, width, and total number of separating plates, respectively. The wall thickness bbv is then the length for heat conduction. If there is any contact or bond resistance present between the fin and tube or plate on the hot or cold fluid side, it is included as an added thermal resistance on the right-hand side of Eq. (3.23) or (3.24). For a heat pipe heat exchanger, additional thermal resistances associated with the heat pipe need to be included on the right-hand side of Eq. (3.23) or (3.24);these resistances are evaporation resistance at the evaporator section of the heat pipe, viscous vapor flow resistance inside the heat pipe (very small), internal wick resistance at the condenser section of the heat pipe, and condensation resistance at the condenser section. If one of the resistances on the right-hand side of Eq. (3.23) or (3.24) is significantly higher than other resistances, it is referred to as the controlling resistance. It is considered significantly dominant when it represents more than 80% of the total resistance. For example, if the cold side is gas (air) and the hot side is condensing steam, the thermal resistance on the gas side will be very high (since h for air is very low compared to that for the condensing steam) and will be referred to as the controlling resistance for that exchanger. However, for a water-to-water heat exchanger, none of the thermal resistances may be dominant if the water flow rates are about the same. The lowest overall thermal resistance in a heat exchanger can be obtained by making the hot- and cold-side thermal resistances about equal (considering wall and fouling resistances is negligible or low). Hence, a low h is often compensated by a high A to make (qOhA)h x (qohA),. This is the reason the surface area on the gas side is about 5 to 10 times higher than that on the liquid-side when the liquid side heat transfer coefficient h is 5 to 10 times higher than the h on the gas side. This would explain why fins are used on the gas sides in a gas-to-liquid or gas-to-phase change exchanger. In Eq. (3.24) or (3.12),the overall heat transfer coefficient Umay be defined optionally in terms of the surface area of either the hot surface, the cold surface, or the wall conduction area. Thus,

HEAT EXCHANGER VARIABLES AND THERMAL CIRCUIT

111

Thus in specifying UA as a product, we don't need to specify A explicitly. However, the option of Ah, A,, or A,v must be specified in evaluating U from the product UA since u h # Uc if Ah # A,. It should be mentioned that the value of R , = I / U A will always be larger than the largest thermal resistance component of Eq. (3.23). This means that UA will always be smaller than the minimum thermal conductance component [a reciprocal of any one term of the right-hand side of Eq. (3.24)]. UA is referred to as the overall thermal conductance. If the overall rate equation is based on a unit surface area

the unit overall thermal resistance is Ro = 1/U. In this case, individual components of resistances are also on a unit area basis, all based on either Ah or A , explicitly as follows:

(3.30~)

(3.30b) where l / U h is the unit overall thermal resistance based on the hot-fluid-side surface area. Similarly, l/Uc is defined. Also Rj = l / ( g h ) j , j= h or c are unit thermal resistances for hot or cold fluids, Rf = l / h f = unit thermal fouling resistance, and R,. = 6,,./k,. = unit wall thermal resistance. For a plain tubular exchanger, qo = 1 ; then from Eq. (3.30), Uo based on the tube outside surface area is given as follows after inserting the value of R,. from the first equation of Eq. (3.26): (3.31~)

(3.31b) Here the subscripts o and i denote the tube outside and inside, respectively; I / U o and 1/U, are the unit overall thermal resistances based on the tube outside and inside surface area, respectively. Knowledge of wall temperature in a heat exchanger is essential to determine the localized hot spots, freeze points, thermal stresses, local fouling characteristics, or boiling/condensing coefficients. In this case, T,v,h and T,.,' can be computed from Eq. (3.22) on a local basis as follows: (3.32)

112

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

Based on the thermal circuit of Fig. 3.4, when R,, is negligible, T,r,h= T,,, = T,,., and Eq. (3.32) reduces to

T,,, =

Th f

[(Rh f Rh,f)/(Rcf Rc,f)l/Tc

+ [(Rh + R h , f ) / ( R c + Rc.f)l

(3.33)

When there is no fouling on either fluid side (Rh,/ = Rc,f = 0 ) , this reduces further to (3.34) Equations (3.32) through (3.34) are also valid for the entire exchanger if all temperatures are used as mean or effective values on respective fluid sides. Example 3.2 In a shell-and-tube feedwater heater, cold water at 15C flowing at the rate of 180 kg/h is preheated to 90C by flue gases from 150C flowing a t the rate of 900 kg/h. The water flows inside the copper tubes (di = 25mm, do = 32mm) having thermal conductivity k,,,= 381 W / m . K. The heat transfer coefficients on gas and water sides are 120 and 1200 W/m2 . K, respectively. The fouling factor on the water side is 0.002m2 . K/W. Determine the flue gas outlet temperature, the overall heat transfer coefficient based on the outside tube diameter, and the true mean temperature difference for heat transfer. Consider specific heats cp for flue gases and water as 1.05 and 4.19J/g. K respectively, and the total tube outside surface area as 5m2. There are no fins inside or outside the tubes, and there is no fouling on the gas side. SOLUTION

Problem Data and Schematic; Fluid flow rates, inlet temperatures, and cold fluid outlet temperature are provided for a shell-and-tube exchanger of prescribed tube inner and outer diameters (Fig. E3.1). Also, the thermal conductivity of the tube and the thermal resistance on the cold fluid side are given. There are no fins on either side of the tubes. Determine: Hot fluid outlet temperature Th,,, overall heat transfer coefficient U , and true mean temperature difference AT,,,. Assumptions: The assumptions invoked in Section 3.2.1 are valid. Hot-fluid-side fouling is negligible.
Flue gas

kuh, cp,h = 1.05 J/g.K

Tc,o = 90C Water T,j = 15C

di= 25 mm
d o = 32 mm it, = 381 W/m.K
Th.o

A,

= 180 kg/h

c ~ ,= , 4.19 J/g.K

hgas=hh=120W/m2.K hwatel= hc=1200W/m*.K = 0.002 m2. WW A,= 5 rn2

kw,f

FIGURE E3.1

HEAT EXCHANGER VARIABLES A N D THERMAL CIRCUIT

113

Analysis: The required heat transfer rate may be obtained from the overall energy balance for the cold fluid [Eq. (3.5)].

- 1 5 ) T = 15,713W l8Okglh (4.19J/g.K)(1000g/kg)(90 (mzq)

Apply the same Eq. (3.5) on the hot fluid side to find the outlet temperature for flue gas:

Since

we get
Th,o = 150C -

15,713W = 90.1"C 0.25 kg/s x 1050J/kg. "C

Ans.

Since U is based on A = Ah = 7rdoLN,, Eq. (3.31) reduces to the following form after substituting the hot-fluid-side fouling factor (l/hf) as zero, and replacing the subscripts o and i of U and h with h and c, respectively.

1 0.032 m[ln(32 mm/25 mm)] 2 x 381 W / m . K 120W/m2. K + 0.032 m 120W/m2 . K x 0.025 m

K/W x 0.032 m + 0.002 m2 .0.025 m

(0.00833 + 0.00001 + 0.00256 + 0.00107) m2 K/W = 0.01 197m2.K/W (69.6%) (0.1%) (21.4%) (8.9%)

Hence,
r/, = 83.54 W/m2 . K Am.

Note that the controlling thermal resistance for this feedwater heater is 69.6% on the flue gas side. Now the mean temperature difference can be determined from Eq. (3.12) as 15,713W AT,,,=-- 4 - 37.6"C UhAh 83.54W/m2. K X 5m2 Ans.

114

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

Discussion and Comments: Since the heat transfer coefficient on the cold side is greater than that on the hot side and the hot- and cold-side surface areas are about the same, the hot side becomes the controlling resistance side. This can be seen from the unit thermal resistance distribution as 69.6% of the total unit thermal resistance on the hot-gas side. The tube wall, made of copper, turned out to be a very good conductor with very small thermal resistance. Notice that the fouling resistance on the water side contributes about one-fifth (21.4%) of the total unit thermal resistance and hence about 21 % surface area penalty. If there had been no fouling on the water side, we would have reduced the heat transfer surface area requirement by about one-fifth. Hence, if it is desired to make a single important improvement to reduce the surface area requirement in this exchanger, the best way would be to employ fins on the gas side (i.e., employing low finned tubing in the exchanger).

3.3 THE E-NTU METHOD

In the E-NTU method, the heat transfer rate from the hot fluid to the cold fluid in the exchanger is expressed as
4 = E C m , n ( T h . i - Tc.!)= G
i n

ATmax

(3.35)

where E is the heat exchanger effectiveness,+ sometimes referred to in the literature as the thermal efficiency, C , , , is the minimum of C h and C,; AT,,, = (Th I - TC,!) is the fluid inlet temperature diference (ITD). The heat exchanger effectiveness E is nondimensional, and it can be shown that in general it is dependent on the number of transfer units NTU, the heat capacity rate ratio C*, and the flow arrangement for a direct-transfer type heat exchanger: E = $(NTU, C*, ffow arrangement) (3.36) Here the functional relationship $ is dependent on the flow arrangement. The three nondimensional groups, E , NTU, and C* are first defined below. The relationship among them is illustrated next.
3.3.1 Heat Exchanger Effectiveness E

Effectiveness E is a measure of thermal performance of a heat exchanger. It is defined for a given heat exchanger of any flow arrangement as a ratio of the actual heat transfer rate from the hot fluid to the cold fluid to the maximum possible heat transfer rate qmax thermodynamically permitted: 4 E=(3.37)
Qmax

'It should be emphasized that the term effectiveness may not be confused with effiiciency. The use of the term efficiency is generally restricted to (1) the efficiency of conversion of energy form A to energy form B, or (2) a comparison of actual system perlormance to the ideal system performance, under comparable operating conditions, from an energy point of view. Since we deal here with a component (heat exchanger) and there is no conversion of different forms of energy in a heat exchanger (although the conversion between heat flow and enthalpy change is present), the term effectiveness is used to designate the efficiency of a heat exchanger. The consequence of the first law of thermodynamics is the energy balance, and hence the definition of the exchanger effectiveness explicitly uses the first law of thermodynamics (see Chapter I 1 for further discussion).

THE 8-NTU METHOD

115

Here it is idealized that there are no flow leakages from one fluid to the other fluid, and vice versa. If there are flow leakages in the exchanger, q represents the total enthalpy gain (or loss) of the Crnin fluid corresponding to its actual flow rate in the outlet (and not inlet) stream. How d o we determine qmax?It would be obtained in a perfect counterjow heat exchanger (recuperator) of infinite surface area, zero longitudinal wall heat conduction, and zero flow leakages from one fluid to the other fluid, operating with fluid flow rates and fluid inlet temperatures the same as those of the actual heat exchanger; also, assumptions 8 to 11, 13, and 14 of Section 3.2.1 are invoked for this perfect counterflow exchanger. This perfect exchanger is the meterbar (or yardstick) used in measuring the degree of perfection of actual exchanger performance. The value of E ranges from 0 to 1. Thus E is like an efficiency factor and has thermodynamic significance. As shown below, in such a perfect heat exchanger, the exit temperature of the fluid with the smaller heat capacity will reach the entering temperature of the larger heat capacity fluid.+ Consider a counterflow heat exchanger having infinite surface area. An overall energy balance for the two fluid streams is

h < C,, (Th,;- Th,o)> ( Tc,o- Tc,i). The temperature drop Based on this equation, for c on the hot fluid side will thus be higher, and over the infinite flow length the hot fluid temperature will approach the inlet temperature of the cold fluid as shown by the two bottom curves in Fig. 3.5, resulting in Th,o = T,,;. Thus for an infinite area counterflow exchanger with C h < C,, we get qmaxas

Similarly, for

c h = C, = C,

Based on Eq. (3.38), for Ch > C,, (Tc,o - T,,;)> (Th,i - Th,o). Hence, Tc,owill approach Th,;over the infinite length, and therefore

Or, more generally, based on Eqs. (3.39) throught (3.41),

where (3.43)

It should be mentioned here that the second law of thermodynamics is involved implicitly in the definition of the exchanger effectiveness since the maximum possible heat transfer rate is limited by the second law. Further discussion of this and related issues is presented in Section 11.2.2.

116

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

FIGURE 3.5 Temperature distributions in a counterflow exchanger of infinite surface area (From Shah 1983).

Exchanger length

Thus q,,, is determined by Eq. (3.42) for defining the measure of the actual performance of a heat exchanger having any flow arrangement. Notice that AT,,, = Th,j - Tc,iin every case and Cminappears in the determination of q,,, regardless of Ch > Cc or
C h

I cc.

Using the value of actual heat transfer rate q from the energy conservation equation (3.5) and qmax from Eq. (3.42), the exchanger effectiveness E of Eq. (3.37) valid for allflow arrangements of the two fluids is given by (3.44) Thus E can be determined directly from the operating temperatures and heat capacity rates. It should be emphasized here that Th,oand Tc,oare the bulk outlet temperatures defined by Eq. (7.10). If there is flow and/or temperature maldistribution at the exchanger inlet, not only the fluid outlet temperatures but also the fluid inlet temperatures should be computed as bulk values and used in Eq. (3.44). An alternative expression of E using q from the rate equation (3.12) and q,,, from Eq. (3.42) is
&=-

U A AT,,, G i n ATmx

(3.45)

Now let us nondimensionalize Eq. (3.7). The mean fluid outlet temperatures T h , o and Tc,o, the dependent variables of Eq. (3.7), are represented in a nondimensional form by the exchanger effectiveness E of Eq. (3.44). There are a number of different ways to arrive

THE e-NTU METHOD

117

at nondimensional groups on which this exchanger effectiveness will depend. Here we consider an approach in which we list all possible nondimensional groups from visual inspection of Eqs. (3.44) and (3.45) as follows and then eliminate those that are not independent; the exchanger effectiveness E is dependent on the following nondimensional groups:

independent

dependent

Note that AT,,, = Th,i - Tc,iin the last three groups of Eq. (3.46) is an independent parameter. In Eq. (3.46), C,,, = C, for Cc > c h and C,,, = c h for c h > C,, so that

(3.47)

In order to show that the third through fifth groups on the right-hand side of Eq. (3.46) are dependent, using Eqs. (3.5) and (3.44), we can show that the first two of the three groups are related as

and using Eq. (3.49, we can show that the fifth group of Eq. (3.46) is

(3.49)
Since the right-hand side of the last equality of Eqs. (3.48) and (3.49) have E , Cmin/Cmax, and UA/Cminas the only nondimensional groups and they are already included in Eq. (3.46), the dimensionless groups of the left-hand side of Eqs. (3.48) and (3.49) are dependent. Thus Eq. (3.46) can be written in terms of nondimensional groups, without a loss of generality, as follows:
E=f$

(-,-

G i n

Cmax

, flow arrangement

= +(NTU, C*, flow arrangement)

(3.50)

where UA/Cmin(number of transfer units = NTU) is a dimensionless parameter under designers control, Cmin/Cmax (heat capacity rate ratio = C*) is a dimensionless operating parameter, and the heat exchanger flow arrangement built into q~ is also a designers parameter. Note that we could have obtained the three nondimensional groups, from the variables and parameters of Eq. (3.7), directly by using Buckinghams 7r theorem (McAdams, 1954) for a given flow arrangement. In Section 11.2, a rigorous modeling approach is presented to determine the same dimensionless groups. A comparison of Eqs. (3.50) and (3.7) will support the advantages claimed for the nondimensional analysis approach. For each flow arrangement, we have reduced a

118

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

seven-parameter problemt [Eq. (3.7)], to a three-parameter problem, [Eq. (3.50)] for a given heat exchanger flow arrangement of two fluids. Before discussing the physical significance of the two independent parameters C* and NTU, let us introduce the definitions of the temperature effectiveness for the hot and cold fluids. The temperature effectiveness &h of the hot fluid is defined as a ratio of the temperature drop of the hot fluid to the fluid inlet temperature difference:

(3.51) Similarly, the temperature effectiveness of the cold fluid is defined as a ratio of the temperature rise of the cold fluid to the fluid inlet temperature difference: (3.52) From an energy balance [Eq. (3.5)], and definitions of &h, and Ch&h= CcEc
E,,

it can be shown that (3.53)

A comparison of temperature effectivenesses with Eq. (3.44) for the exchanger (heat transfer) effectiveness reveals that they are related by
& = - EC hh =

for
E~ /C*

c h = Cmi,

Cmin
c c

for Ch = C,,, for C, = Cmin

(3.54)

&=-

& ,

C m a x

Ec

&,/C*

for C, = C,,,

(3.55)

Now let us define and discuss C* and NTU.

3.3.2 Heat Capacity Rate Ratio C*

C* is simply a ratio of the smaller to larger heat capacity rate for the two fluid streams so that C* 5 1. A heat exchanger is considered balanced when C* = 1:
cmin c* --=-cmax

(~cp)min (~Cp)max

tTc,o - Tc,i)/(Th,i- Th,o) (Th,;- Th,o)/(Tc,o- TC,;)

for c h = Cmin for

cc= Cmin

(3.56)

C* is a heat exchanger operating parameter since it is dependent on mass flow rates and/or temperatures of the fluids in the exchanger. The C,,, fluid experiences a fluid in the absence smaller temperature change than the temperature change for the Cmin of extraneous heat losses, as can be found from the energy balance:
q=c h AT, = c, AT,
In counting, we have considered only one dependent variable of Eq. (3.7).

(3.57)

THE E-NTU METHOD

119

where the temperature ranges AT, and AT, are

Let us reemphasize that for a condensing or evaporating fluid at ideally constant temperature, the A T range (rise or drop) is zero, and hence the heat capacity rate C approaches infinity for a finite q = C AT. Since C = ritcp, the effective specific heat of as condensing or evaporating fluid is hence infinity. The C* = 0 case then represents Cmin finite and C,, approaching infinity in the sense just discussed. The C* = 0 case with Cmin = ( r i t ~ , , )= ~0 ~~ is not of practical importance, since rit = 0 in that case and hence there is no flow on the Cmin side of the exchanger.

3.3.3 Number of Transfer Units NTU

The number oftransfer units NTU is defined as a ratio of the overall thermal conductance to the smaller heat capacity rate: (3.59) If U is not a constant, the definition of the second equality applies. NTU may also be interpreted as the relative magnitude of the heat transfer rate compared to the rate of enthalpy change of the smaller heat capacity rate fluid. A substitution of the U A magnitude from Eq. (3.24) into Eq. (3.59) for U as constant results in

NTU designates the nondimensional heat transfer size or thermal size of the exchanger, and therefore it is a design parameter. NTU provides a compound measure of the heat exchanger size through the product of heat transfer surface area A and the overall heat transfer coefficient U. Hence, in general, NTU does not necessarily indicate the physical size of the exchanger. In Lontrast, the heat transfer surface area designates the physical size of a heat exchanger. A large value of NTU does not necessarily mean that a heat exchanger is large in physical size. As a matter of fact, the automotive gas turbine regenerator at the idle operating point may have NTU x 10 and core volume V x 0.01 m3, whereas a chemical plant shell-and-tube exchanger may have NTU x 1 and V x loom'. However, when comparing heat exchangers for a specijic application, UIC,,, approximately remains constant; and in this case, a higher NTU value means a heat exchanger larger in physical size. Hence, NTU is sometimes also referred to as a heat exchanger sizefactor. In general, higher NTU is obtained by increasing either U or A or both or by decreasing C,,,. Whereas a change in C,,, affects NTU directly, a change in C,, (i.e., its flow rate) affects h on the C,, side. This in turn influences U and NTU. Thus, a change in the value of C* may have direct or indirect effect on NTU. NTU is also variously referred to as a performance factor or thermal length 0 in the plate heat exchanger' literature, and as reduced thermal fzux in the shell-and-tube
'In a PHE with chevron plates, a plate with high chevron angle B has a high heat transfer coefficient and high pressure drop in general and is referred to as a hardp/ate; in contrast, a s o f f p l a f ehas a low chevron angle 4 and low heat transfer coefficient and low pressure drop in general.

120

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

exchanger literature. Other names are given in the following text with appropriate interpretations. At low values of NTU, the exchanger effectiveness is low. With increasing values of NTU, the exchanger effectiveness generally increases, and in the limit, it approaches a thermodynamic asymptotic value. Note that the perfect exchanger requires that AT,,,. The following approximate values NTU -+ 3= (because A + 00) for qmax= Cmin of NTU will illustrate this point further. Automobile radiator: Steam plant condenser: Regenerator for industrial gas turbine engine: Regenerator for Stirling engine: Regenerator for an LNG plant: N T U x 0.5 + E x 40% NTUX 1 + ~ ~ 6 3 % N T U x 10 + E x 90% N T U x 50 + E x 98% N T U x 200 + E x 99%

Another interpretation of N T U as nondimensional residence time is as follows by substituting Cmin = Cmin/rd in the definition of NTU:

(3.61)
Here R, = l/UA is the overall thermal resistance; Cmin =( M C ~ ) =, c,inrd ~ ~ is the minimum-side fluid heat capacitance ( M = h i d mass in the exchanger at an instant of time) in the exchanger at any instant of time; and Td is the dwell time, residence time, or transit time of a fluid particle passing through the exchanger. Thus, N T U can be interpreted as a nondimensional residence time or a ratio of residence time to the time constant of the Cmin fluid in the exchanger at an instant. NTU might well be expected to play an important part in the transient problem! And from the viewpoint of an observer riding with a particle in the Cminstream, he or she would indeed have a transient temperature-time experience. Yet another interpretation of NTU is as follows. NTU is related to A T , from Eq. (3.45) with the use of Eq. (3.44) as

Therefore, NTU is referred to as a temperature ratio (TR), where

is a maximum of AT, and AT,. When A T , Thus, NTU = AT,a,>i/AT,, where ATmax,i is equal to A T , or AT, whichever is larger, N T U = 1. Notice that Eq. (3.63) is convenient for small values of NTU, in which case A T , x T,,h - T,,, is a good approximation; there is no need to calculate AT,, or F (see Section 3.7). Here Tm,h and T,,,, are the arithmetic mean values of the corresponding terminal temperatures.

EFFECTIVENESS-NUMBER OF TRANSFER UNIT RELATIONSHIPS

121

NTU is also directly related to the overall (total) Stanton number St, formulated with

U in place of h of Eq. (7.28) as

4L NTU = St, Dh (3.64)

Thus, NTU can also be interpreted as a modified Stanton number. Note that here the hydraulic diameter Dh is defined as follows depending on the type of heat exchanger surface geometry involved.

Dh=(

4 x flow area - 4A0 - 4A0L wetted perimeter P A 4 x core flow volume = 4pV 4p =-=fluid contact surface area A p

4p
a

(3.65)

where p is the porosity, a ratio of void volume to total volume of concern. Here the first definition of Dh is for constant cross-sectional flow passages in a heat exchanger. However, when flow area is expanding/contracting across flow cross sections along the flow length as in three-dimensional flow passages (such as in a corrugated perforated fin geometry of Fig. 1.29f), the more general second definition is applicable. In the second definition, Dh = 4p/p for plate-fin type and regenerative surfaces; for tube bundles and tube-fin surfaces, Dh = 4p/a. Note that heat transfer and pressure drop Dh magnitudes will be different if the heated and flow friction perimeters are different, as illustrated in the footnote of p. 9. Equations (3.63) and (3.59) may also be interpreted as the number of transfer units required by the heat duty (NTU = ATm,,,i/AT,) and the number of transfer units achieved by the heat exchanger (NTU = UA/Cmin),respectively. The foregoing definitions and interpretations are for the overall NTU for the exchanger. The number of heat transfer units individually on the hot and cold sides of the exchanger may be defined as: (3.66) We use ntuh and ntu, in Chapter 9 when outlining the solution procedure for the sizing problem for extended surface heat exchangers. The overall thermal resistance equation (3.24), in the absence of fouling resistances, can then be presented in terms of overall and individual number of transfer units as (3.67)

3.4 EFFECTIVENESS-NUMBER OF TRANSFER UNIT RELATIONSHIPS In the preceding section we demonstrated that E is a function of NTU, C*, and flow arrangement. We now derive this functional relationship for a single-pass counterflow

122

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

m

Th,I

I I

1 1

--h,II

exchanger and then summarize similar functional relationships for single-pass and multipass flow arrangements.

3.4.1 Single-Pass Exchangers

3.4.1.1 CounterJlow Exchanger.+ Consider a counterflow heat exchanger with the tem-

perature distributions for hot and cold fluids as shown in Fig. 3.6. The fluid temperatures on the left-hand end of the exchanger are denoted with the subscript I, and those on the other end with the subscript 11. In the analysis, we consider the overall counterflow exchanger as shown in Fig. 3.3 with only two passages. This is because the idealizations made in Section 3.2.1 (such as uniform velocities and temperatures at inlet, uniform surface area distribution, uniform U , etc.) are also invoked here. Thus, in fact, the hot-fluid passage shown represents all hot-fluid flow passages, and the cold-flow passage shown consists of all cold-fluid flow

' We derive the heat exchanger effectivenessin this section, and show how to obtain temperature distributions in
Chapter I 1.

EFFECTIVENESS-NUMBER OF TRANSFER UNIT RELATIONSHIPS

123

passages. This is the reason that c h and C, are associated with the respective single-flow passages shown in Fig. 3.3 and not dch and dCc. Based on an energy balance consideration on the differential element dx,
dq = -ch dTh = -C, dTc

(3.68)

Here Th represents the bulk temperature of the hot fluid in the differential element dx of Fig. 3.6. T, is defined in a similar manner for the cold fluid. The rate equations applied individually to the dx length elements of the hot fluid, wall, and the cold fluid yield

dq =

{ S,7
k , A,dx

(%hA)k(Th

-Tw,h)y

dx

for the hot fluid for the wall for the cold fluid

(3.69) (3.70) (3.71)

( Tw,h- T&,,,) dx

(T&~),(T,,, - T,)

Note that considering a general case, we have included the extended surface efficiency 71, in Eqs. (3.69) and (3.71), although fins are not shown in Fig. 3.6. After getting the expressions for individual temperature differences from Eqs. (3.69)-(3.7 l), adding them up, and rearranging, we get
dx dq = uA(Th - T c ) L

(3.72)

Here U represents the local overall heat transfer coefficient for the element dA or dx. However, we treat this local U the same as U,,, for the entire exchanger, and hence U will be treated as a constant throughout the exchanger. Hence UA in Eq. (3.72) is given by Eq. (3.20) or (3.24) without the fouling resistances on the hot and cold sides. If the fouling or other resistances are present, UA in Eq. (3.72) will have those resistances included. Elimination of dq from Eqs. (3.68) and (3.72) will result in two ordinary differential equations. We need two boundary conditions to solve them and they are
Th(X = 0 ) = Th,, T,(x = L ) = T,,,,

(3.73)

Now let us solve the set of Eqs. (3.68), (3.72), and (3.73) as follows to obtain a ratio of terminal temperature differences so that we can determine the exchanger effectiveness directly. We derive the temperature distributions in Section 1 1.2.1. Here, effectively, we will solve one differential equation that is based on the energy balance and rate equations. Substituting the values Of dTh and dT, in terms of dq, Eq. (3.68) can be rearranged to
d ( T h - T,) = - - ( A c

dq

(3.74)

:h)

Eliminating dq from Eqs. (3.72) and (3.74) and rearranging yields

d(Th Th

- Tc) - Tc

= - (1 -

2)

uA dx c h L

(3.75)

124

BASIC THERMAL DESlGN THEORY FOR RECUPERATORS

Integrating this equation from the hot-fluid inlet (section I) to the outlet section (section 11) yields (3.76) It can be shown algebraically that the left-hand side of Eq. (3.76) is

Now employ the definitions of the temperature effectivenesses from Eqs. (3.5 1) and (3.52), using the nomenclature of Fig. 3.6: (3.78) Substituting the definitions of and
E,

in Eq. (3.77), it reduces to (3.79)

where Eq. (3.53) is used to obtain the last equality. Substituting Eq. (3.79) into (3.76) and rearranging, we get
1 - exp[-(UA/Ch)(l
"I

= 1 - (Ch/Cc)exp[-(UA/Ch)(l

- ch/cc)l - ch/cc)]

(3.80)

Now the temperature effectiveness of the cold fluid, E,, can be determined either directly by employing its definition [Eq. (3.52)], or by substituting Eq. (3.80) into (3.53). Using the second approach, gives us
(3.81)

Notice that the argument of the exponential terms has been rearranged. By multiplying both numerator and denominator by exp {-(UA/C,)[l - ( C c / C h ) ]and } rearranging, we get
E,

1 - exp[-(UA/Cc)(1 - Cc/C,)l 1 - (Ce/Ch)exp[-(UA/Cc)(l - cc/ch)l

(3.82)

A comparison of Eqs. (3.80) and (3.82) reveals that Eq. (3.82) can be obtained directly from Eq. (3.80) by replacing the subscripts h with c and c with h. To generalize the solution, let Cmi, = Cc, c*= c,/ch and NTU = UA/Cmi, = UA/Cc.In this case, E = E, from Eq. (3.52), and Eq. (3.82) reduces to
E

= EcJ =

1 - exp[-NTU(1 - C*)] 1 - C* exp[-NTU( 1 - C*)]

(3.83)

EFFECTIVENESS-NUMBER OF TRANSFER UNIT RELATIONSHIPS

125

However, if Cmin= ch, then C* = C h / c , , NTU = UA/Cmin= UA/Ch. In this case, = &h from Eq. (3.51), and Eq. (3.80) reduces to the exchanger effectiveness expression of Eq. (3.83). Thus, regardless of whichfluid has the minimum heat capacity rate, the ENTU expression for the counterflow exchanger is given by Eq. (3.83). Two limiting cases of interest of Eq. (3.83) are C* = 0 and 1. For the special case of C* = 0 (an evaporator or a condenser), the exchanger effectiveness E, Eq. (3.83), reduces to
E

1 - exp(-NTU)

(3.84)

Note that when C* = 0, the temperature of the C,, fluid remains constant throughout the exchanger, as shown in Fig. 3 . l b and c. In this case, the Cmin fluid can have any arbitrary flow arrangement. Hence Eq. (3.84) is valid for all flow arrangements when c*= 0. For the special case of C* = 1, Eq. (3.83) reduces to the OjO form. Hence, using 1Hospitals rule (taking the derivatives of the numerator and the denominator separately with respect to C* and substituting C* = 1 in the resultant equation), we get NTU =I+NTU For all 0 < C* < 1, the value of shown in Fig. 3.1.
100
E

(3.85)

falls in between those of Eqs. (3.84) and (3.85), as

ao

60

8
0

40

Heat transfer sutface

20

3 NTU

FIGURE 3.7 Counterflow exchanger E as a function of NTU and C*.

124

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

By inverting Eq. (3.83), N T U for a counterflow exchanger can be expressed explicitly as a function of E and C*: 1
~

NTU =-In

1 - C*E

1-c*

I--

(3.86)

which for C* = 0 and C* = 1 (using 1Hospitals rule) are given by For C* = 0: NTU = In1 I- or NTU = I n
Th.i Th.1 31

Tc.0

for T ~=,constant ~
(3.873 (3.88)

The E-NTU results for the counterflow exchanger are presented in Fig. 3.7. The following important observations may be made by reviewing Fig. 3.7:
1. The heat exchanger effectiveness E increases monotonically with increasing values of NTU for a specified C*. For all C*, E -+ 1 as NTU -+ w. Note that this is true for the counterJow exchanger, and E may not necessarily approach unity for many other exchanger flow arrangements, as can be found from Tables 3.3 and 3.6. 2. The exchanger effectiveness E increases with decreasing values of C* for a specified

NTU. 3. For E 6 40%, the heat capacity rate ratio C* does not have a significant influence on the exchanger effectiveness c. As a matter of fact, it can be shown that when NTU -+ 0, the effectiveness for all flow arrangements reduces to
E

! z NTU[I - I N T U ( 1

+ c*)]

(3.89)

This formula will yield highly accurate results for decreasing value of NTU for NTU 6 0.4. 4. Although not obvious from Fig. 3.7, but based on Eq. (3.84), the E vs. NTU curve is identical for all exchanger flow arrangements, including those of Tables 3.3 and 3.6 for C* = 0. 5. Heat exchanger effectiveness increases with increasing NTU as noted above in item 1, but at a diminishing rate. For example, increasing NTU from 0.5 to 1 at C* = 1 increases E from 0.333 to 0.50; a 50% increase in E for a 100% increase in NTU (or approximately 100% increase in surface area). Increasing NTU from 1 to 2 at C* = 1 increases E from 0.50 to 0.667, a 33% increase in E for a 100% increase in NTU (or size). Increasing NTU from 2 to 4 at C* = 1 increases E from 0.667 to 0.8, a 20% increase in E for a 100% increase in NTU (or size). This clearly shows a diminishing rate of increase in E with NTU. 6. Because of the asymptotic nature of the E-NTU curves, a significant increase in NTU and hence in the exchanger size is required for a small increase in E at high values of E . For example, for C* = 1, E = 90% at NTU = 9, and E = 92% at NTU = 11.5. Thus an increase of 2 percentage points in E requires an increase

EFFECTIVENESS-NUMBER OF TRANSFER UNIT RELATIONSHIPS

127

of 28% in N T U and hence in the exchanger size for the same surface area and flow rates. Alternatively, a larger increase in N T U (or the size of the exchanger) is required to compensate for the same (or small) amount of heat loss to the surroundings at high values of E in comparison to that for a lower E exchanger. The counterflow exchanger has the highest exchanger effectiveness E for specified NTU and C* of that for all other exchanger flow arrangements. Thus, for a given NTU and C*, maximum heat transfer performance is achieved for counterflow; alternatively, the heat transfer surface area is utilized most efficiently for counterflow compared to all other flow arrangements. Should we design heat exchangers at high effectiveness for maximum heat transfer? Let us interpret the results of Fig. 3.7 from the industrial point of view. It should be emphasized that many industrial heat exchangers are not counterflow, and the following discussion in general is valid for other flow arrangements. When the heat exchanger cost is an important consideration, most heat exchangers are designed in the approximate linear range of E-NTU curves (NTU 5 2 or E 5 60%, such as in Fig. 3.7) that will meet the required heat duty with appropriate values of C,,, and AT,,,. The reason for this is that an increase in the exchanger size (NTU) will increase with E approximately linearly and hence will result in a good return on the investment of added surface area. However, when the exchanger is a component in the system, and an increase in the exchanger effectiveness has a significant impact on reducing the system operating cost compared to an increase in the exchanger cost, the exchangers are designed for high effectivenesses. For example, a 1 % increase in the regenerator effectiveness approximately increases about 0.5% the thermal efficiency of an electricity generating gas turbine power plant. This increase in the power plant eficiency will translate into millions of dollars worth of additional electricity generated annually. Hence, the cost of the regenerator represents only a small fraction of the annual savings generated. Therefore, most gas turbine regenerators are designed at about E = 90%. A 1 % increase in the effectiveness of a cryogenic heat exchanger used in an air separation plant will decrease the compressor power consumption by about 5%. This will translate into an annual savings of more than $500,000 (in 1995 dollars) in operating costs for a 1000-ton oxygen plant. That is why most cryogenic heat exchangers are designed for E = 95%.
3.4.1.2 Exchangers with Other Flow Arrangements. The derivation of E-NTU formulas for other heat exchanger flow arrangements becomes more difficult as the flow arrangement becomes more complicated (Sekulit et al., 1999). The solutions have been obtained in the recent past for many complicated flow arrangements, as in Table 3.6 and Shah and Pignotti (1989). Temperature distributions for counterflow, parallelflow, and unmixed-unmixed crossflow exchangers are derived in Sections 1 1.2.1 and 11.2.4.

E-NTU Results. Table 3.3 and Figs. 3.8 through 3.11 show the E-NTU formulas and results for some heat exchanger flow arrangements. NTU can be expressed explicitly as a function of E and C* only for some flow arrangements and those formulas are presented in Table 3.4. For all other flow arrangements, N T U is an implicit function of E and C* and can be determined iteratively o r by using appropriate analytical methods to solve the equation f(NTU) = 0. C* cannot be expressed explicitly as a function of E and NTU for any exchanger flow arrangements.

t W . 2

TABLE 3.3 E-NTU Formulas and Limiting Values of E for C = 1 and NTU
Flow Arrangement E-NTU Formulas

w for Various Exchanger Flow Arrangements

E-NTU Formulas for C* = 1
=-

Asymptotic Value o fE When NTU + 03

%Counterflow
. . .a , . .
E=

1 - C* exp[-NTU( 1 - C*
1 - exp[-NTU(1 1+C*

I
E =

+ NTU
- exp( -

NTU

E =

I fop all C*
I

'tf
4

Parallelflow . . . . .+ Crossflow, both fluids unmixed

+ C')]

f [I

mu)]

=-

I +C*
1 for all C*

E =

- exp(-NTU)

-exp[-( I

+ C*)NTU]

03

Same as general formula with C* = 1 C*"P,,(NTU)

E =

"=I

Crossflow, one fluid mixed, other unmixed

For C,," mixed, C,,, unmixed,

E

= 1 - exp{-[I - exp(-NTU)]}

For C , , , mixed, E = I - exp(-I/C*)

= 1 - exp{-[I - exp(-NTU . C*)]/C*)

...@..

For C,, =-(I C*

1

mixed, Cmin unmixed, -exp{-C*[l -exp(-NTU)]})

E =

- exp{-[I

exp(-NTU)]}

For C , , , mixed, E = [ I - exp(-C*)]/C*

1-2 shell-and-tube exchanger; shell fluid mixed; TEMA E shell

2
=

+ C*) + ( I + C*2)"2coth(r/2) where r = NTU(I + C*')'/'

(1

2 2+dcoth(r/2)

E=

2 (I

+ C*) + ( 1 + C*2)'/2

coth(r/2)

(I

+ c - ~ ) / (I - c - ~ )

where r = d N T U

EFFECTIVENESS-NUMBER OF TRANSFER UNIT RELATIONSHIPS

100

/
80

c* = 0.0
0.2
0.4

Heat transfer surface

129

60

0.6
0.8 1.o

s
0

40

20

I
0
1 2
C' = 0.0
4

3
4

0
5

NTU

FIGURE 3.8 Parallelflow exchanger E as a function of NTU and C*.

100

80

60

ae 0
40

20
I

1 Unmixed fluid
I I I I I I I I I
1 1

0
0

3 NTU

FIGURE 3.9 Unmixed-unmixed crossflow exchanger E as a function of NTU and C*.

130

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

100

80

60

s
W

40

20

0
0
1

3
NTU

FIGU RE 3.10 Unmixed-mixed crossflow exchanger E as a function of NIU and C*.

100

80

60

s
W

40

20

0
0
1

3
NTU

FIGURE 3.11 Mixed-mixed crossflow exchanger E as a function of NTU and C*

EFFECTIVENESS-NUMBER O F TRANSFER UNIT RELATIONSHIPS

131

TABLE 3.4 NTU as an Explicit Function of E and Ct for Known Heat Exchanger Flow Arrangements

Flow Arrangement Counterflow

Formula 1 NTU =-In 1-c* NTU = 1- E

E

l-C*E - (C* < 1)

I-

(C* = 1)

Parallel flow

NTU = -

In[l -- E ( 1 + C*)] 1 + c*

Crossflow (single pass)

,,C ,

(mixed), Cmin (unmixed) (unmixed)

NTU = -In 1 + -In(

1 C*

[ d*

1 - C*E)

C,in (mixed),,,C ,

NTU = - -In[1 + C* In( 1 - E ) ] 1 2-41+C*-D) NTU =--In D 2-41+C*+D) where D = (1 + C*2)/2

1-2 TEMA E Shell-and-Tube

All exchangers with C* = 0

NTU = - In( 1 - E )

From a review of Fig. 3.1 1, it is found for C* > 0 that the effectiveness of the mixedmixed crossflow exchanger increases with increasing NTU up to some value of NTU beyond which increasing NTU results in a decrease in E . The reason for this unexpected behavior is that due to the idealized complete mixing of both fluids, the hot-fluid temperature is lower than the cold-fluid temperature in a part of the exchanger, resulting in a temperature cross (discussed in Section 3.6.1.2) and existence of entropy generation extrema (discussed in Section 11.4). Hence, heat transfer takes place in the reverse direction, which in turn reduces the exchanger effectiveness.
Interpretation of E-NTIJ Results. As noted in Eq. (3.7), the heat exchanger design problem in general has six independent and one or more dependent variables. For analysis and concise presentation of the results, we arrived ict a total of three dimensionless group; ( E , NTU, and C*) for this method. However, from the industrial viewpoint, the final objective in general is to determine dimensional variables and parameters of interest in any heat exchanger problem. The nondimensionalization is the shortest road to obtaining the results of interest. In this regard, let us review Eq. (3.35) again:

3.4.1.3

4 = E C m i n ( T h , i - Tc,i) = E(hCp)rnin(Th,i- Tc,i)

(3.90)

If we increase the flow rate on the C, = Cmi,side (such as on the air side of a water-to-air reduce NTU (= UA/C,i,), and heat exchanger), it will increase C* (= Crnin/Crnax),

Next Page

132

BASIC THERMAL DESIGN THEORY FOR RECUPERATORS

hence will reduce E nonlinearly at a lower rate (see Figs. 3.7 to 3.11). Thus from Eq. (3.90), q will increase linearly with increasing Cminand q will decrease nonlinearly at a lower rate with decreasing E . The net effect will be that q will increase. From Eq. (3.9, (3.91) Thus, overall, a lower increase in q and a linear increase in C, will yield reduced (Tc,o - T,,;);it means that Tc,owill be lower for a given T,,i. Again from Eq. (3.91), T/l,o will also be lower with increased q for given c h and TI^,^. These results are shown qualitatively by an arrow up or down in Table 3.5 depending on whether the particular value increases or decreases. It is interesting to note that both Th+ and T,+ in this table can either increase, decrease, or one increases and the other decreases! A thorough understanding of the qualitative trends of the variables of this table is necessary to interpret the experimental/analytical results and anomalies or unexpected test results. Reviewing Eq. (3.90), the desired heat transfer rate q in a heat exchanger can be obtained by varying either E of Cminsince the ITD (= Th,i - T,,;) is generally a fixed design value in many applications. The effectiveness E can be varied by changing the surface area A (the size of the exchanger) and hence NTU, impacting the capital cost; E increases with A or NTU nonlinearly and asymptotically. The minimum heat capacity (= h c p ) can be varied by changing the mass flow rate through a fluid pumping rate Cmin device (fan, blower, or pump); the change in Cmin has a direct (linear) impact on q. Hence, in general, specified percentage change in Cmin will have more impact on q than the change in E due to the same percentage change in A or the size of the exchanger. Current practice in many industries is to maintain the heat exchanger effectiveness 60% or lower to optimize the heat exchanger cost. The desired heat transfer rate is then obtained by using a proper capacity fluid pumping device, which in turn has a major impact on the operating cost. Thus the heat exchanger and the fluid pumping device are selected individually. However, to get the desired heat transfer rate q in an exchanger, the better cost effective approach would be to select optimally the heat exchanger size (NTU) and the fluid pumping power to get the appropriate flow rate (and hence Cmin). Also, a review of Eq. (3.90) indicates that if there is an insufficient fluid flow rate m, the exchanger will not provide the desired performance even if E + 100% for a given

TABLE 3.5 Effect of Increasing One Independent Variable at a Time on the Remaining Heat Exchanger Variables Based on the e-NTU Relationships'

Specific Variable with Increasing Value

mh or
Th,i
Tc,i

Variables Affected NTU

C*

Th.o

T C . 0

c h

mc or C,

t 1
-

I
T
-

I
-

t t t

I
t t

h, or A,
hh

or Ah
= C, for this table.

t t

t 1 t T 1 1

t 1 t t t

'C,,,

T, increase; -, no change; 1, decrease